Social networks are increasingly being recognized as an important determinant of health behaviors, including use of marijuana and other drugs among adolescents. While knowledge about how social networks influence behaviors is expanding rapidly, translating these findings into actionable health interventions or policies is difficult. However, the school environment may be a potentially modifiable factor that has a strong impact on social networks and on adolescent health behaviors. In this proposed study, we will take advantage of a natural experiment and examine the impact of successful public charter schools on social networks and marijuana use. Charter schools have rapidly multiplied in the U.S., primarily in low-income communities. While not all have been successful, many, particularly those run by larger charter management organizations, have been effective in raising academic performance. Two of the largest organizations, Green Dot and Alliance, have more than 30 schools in Los Angeles, serving several thousand low-income, minority students annually. These schools have graduation rates of 85-100% compared to 50% at nearby public schools, and standardized tests scores 2.5 standard deviations higher than nearby public schools. Despite the clear impact of these schools on students' academic achievement, the impact on social networks and health behaviors has not been studied. Because admission to these schools is based on a random lottery that ignores demographics or previous academic performance, Green Dot and Alliance schools provide a perfect setting in which to conduct a natural experiment. We propose a 4-year longitudinal study, sampling among 8th grade students who apply to one of several Green Dot or Alliance charter schools. We will compare those who are randomly chosen for admission to the school (Experimental Group) with those who are not selected for admission (Control Group). We will follow the natural experimental cohort over 4 years exploring how social networks form and evolve over time and how they are related to marijuana use and other risky health behaviors in two very different academic environments. We will examine the impact of high-performing charter high schools on student's social networks and on use of marijuana and other risky behaviors (alcohol, tobacco and other drug use, sexual activity, and violence). We will also examine whether social networks acts as a mediator in the impact of these schools on marijuana use and other risky behaviors. If successful schools influences marijuana use as hypothesized, the results of the proposed study will provide extraordinary and compelling evidence that not only can education reform be achieved, but also that such reform has a wide range of benefits to society including better health. These results could stimulate new research and efforts on ways to improve health through education reform and also through changing social networks.